Evaluation of CA-125, HE4, ROMA, and Copenhagen Index for Preoperative Risk Stratification of Adnexal Masses in Indian Women: A Prospective Cross-Sectional Study

Technical Summary: Evaluation of CA-125, HE4, ROMA, and Copenhagen Index for Preoperative Risk Stratification of Adnexal Masses in Indian Women

Problem Statement
Ovarian cancer represents a significant global and national health burden, accounting for approximately 207,000 deaths annually worldwide and contributing roughly 16% of the global mortality burden in India. While ultrasonography is the cornerstone for evaluating adnexal masses, access to expert interpretation and standardized scoring systems (such as IOTA or RMI) is often variable in low-resource and high-volume healthcare settings. Serum CA-125, the most widely used biomarker, has limitations in specificity, particularly in premenopausal women where benign conditions like endometriosis and pelvic inflammatory disease can cause false-positive elevations. Furthermore, the high prevalence of genital tuberculosis in India further complicates the differential diagnosis, as tubercular masses can also elevate CA-125. There is a need to evaluate newer biomarkers (HE-4) and risk algorithms (ROMA, Copenhagen Index) to determine their utility as adjunctive triage tools for distinguishing benign from malignant adnexal masses in the Indian population, specifically to facilitate timely referral to gynecologic oncology services.

Methodology
This was a single-center, prospective cross-sectional study conducted at the All India Institute of Medical Sciences (AIIMS), New Delhi, between December 2021 and December 2022. The study enrolled 101 women presenting with adnexal masses of suspected ovarian origin; after exclusions for incomplete records or uncertain diagnoses, 86 participants were included in the final analysis.

• Exclusion Criteria: Pregnant women, those with prior bilateral oophorectomy, and non-consenting patients.
• Data Collection: Serum levels of CA-125 and Human Epididymis Protein 4 (HE-4) were measured using Chemiluminescence Microparticle Immunoassay (CMIA).
• Calculations: The Risk of Ovarian Malignancy Algorithm (ROMA) and the Copenhagen Index (CPH-I) were calculated based on the biomarker levels, age, and menopausal status (for ROMA).
• Statistical Analysis: Diagnostic accuracy was assessed using sensitivity, specificity, positive/negative predictive values, and likelihood ratios. Receiver Operating Characteristic (ROC) curves were plotted to determine the Area Under the Curve (AUC). Optimal cutoffs were derived using Youden's J statistic. Univariable and multivariable logistic regression analyses were performed to identify predictors of biomarker misclassification. Borderline ovarian tumors were excluded from primary diagnostic accuracy calculations but analyzed descriptively.

Key Contributions

1. Context-Specific Validation: The study provides data on the performance of these biomarkers specifically within the Indian demographic, where the prevalence of endometriosis and genital tuberculosis creates a unique diagnostic challenge for CA-125 specificity.
2. Comparative Analysis: It offers a direct comparison of CA-125, HE-4, ROMA, and CPH-I across the entire cohort and stratified by menopausal status (premenopausal vs. postmenopausal).
3. Discrepancy Analysis: The study specifically investigates the causes of false-positive and false-negative results, identifying endometriosis and tubercular masses as significant predictors of CA-125 discrepancies.
4. Algorithm Performance: It evaluates the practical utility of CPH-I, which relies on age rather than menopausal classification, potentially simplifying its application in clinical settings where menopausal status may be ambiguous.

Results

• Cohort Characteristics: Of the 86 participants, 58 (67%) had benign masses, 23 (27%) had malignant tumors, and 5 (6%) had borderline ovarian tumors. Malignant cases were significantly older (median 45 years) compared to benign cases (median 30 years).
• Overall Performance: In the overall cohort, ROMA demonstrated the highest diagnostic accuracy (AUC: 0.954), followed closely by HE-4 (AUC: 0.953) and CPH-I (AUC: 0.940). CA-125 showed the lowest AUC (0.895).
• Premenopausal Subgroup: HE-4 and ROMA demonstrated superior performance with AUCs of 0.938 and 0.942, respectively. Both achieved 100% specificity and 75% sensitivity. In contrast, CA-125 showed lower specificity (69.2%) despite similar sensitivity.
• Postmenopausal Subgroup: The Copenhagen Index demonstrated the highest diagnostic accuracy (AUC: 0.924) and specificity (94.83%). CA-125 and ROMA showed high sensitivity (100%) but lower specificity compared to CPH-I.
• Optimal Cutoffs: Study-derived optimal cutoffs using Youden's index were 29.6 U/mL for CA-125, 5.3% for CPH-I, and specific values for HE-4 and ROMA based on menopausal status.
• Discrepancy Analysis: CA-125 had the highest frequency of discordant classifications (21 cases), primarily driven by false positives in benign inflammatory conditions. Endometriosis and tubercular masses were identified as independent predictors of CA-125 discrepancy (Multivariable OR: 3.30 and 14.2, respectively). HE-4-based algorithms had fewer discrepancies, though misclassifications occurred in rare germ-cell tumors (dysgerminoma, mixed GCT).
• Borderline Tumors: Four out of five borderline ovarian tumors were categorized as "low risk" by both ROMA and CPH-I, suggesting these indices may be less likely to overcall borderline lesions as invasive malignancy compared to CA-125.

Significance and Claims
The paper concludes that while CA-125 remains a widely accessible and sensitive frontline biomarker, its specificity is significantly compromised in the presence of benign inflammatory conditions common in India, such as endometriosis and genital tuberculosis.

• Premenopausal Utility: HE-4 and ROMA offer improved specificity over CA-125 for premenopausal women, making them valuable for triaging patients who require referral to gynecologic oncology.
• Postmenopausal Utility: The Copenhagen Index demonstrates the highest specificity and AUC in postmenopausal women. Its reliance on age rather than menopausal classification offers a practical advantage in clinical workflows.
• Clinical Role: The authors emphasize that these biomarkers should serve as adjunctive triage tools alongside imaging and clinical assessment, rather than standalone diagnostic tests. Their primary value lies in facilitating timely referral to specialized care, particularly in resource-limited settings where advanced imaging expertise may not be uniformly available.
• Caveats: The authors explicitly state that the findings must be interpreted cautiously due to the single-center design, relatively small sample size, and limited number of borderline tumor cases. They call for larger, multicentric studies incorporating imaging-based scoring systems and health-economic analyses to validate these results.